Fashion Facebook ad examples

Misdiagnosis / Reframe is the right family for the single most-felt and least-advertised problem in womenswear, and the reframe is what makes it publishable. The reader owns a size 8, a 10, a 12 and a 14 of the same trouser and has quietly concluded the problem is her. The ad moves the blame onto grading that drifts every season, which is both true and — importantly — a claim about the industry rather than about her body. That distinction is the whole design rule of this page: the same insight written as "if your waist has changed" would be an assertion about the reader and inside Meta's personal attributes policy.

The picture is the argument, which is rare and worth stealing. Four identical trousers on a rail tagged 8, 10, 12 and 14, one tape measure running across all four waistbands reading 30 in at every stop, and a caption that just says "Same waist. Four labels." There is no model, no styling and nothing to interpret. It is also the only ad in the batch whose 1.91:1 export fills the frame properly — a diagram survives a wide crop because it has nothing in the top and bottom margins to lose, which is exactly why the two padded exports elsewhere on this page are photographs.

Two things before it runs. The rulers rendered along the top and bottom edges do not count correctly — the top scale reads 0, 10, 20, 30, 40, 60, and the bottom repeats a number mid-sequence — which is decoration on a diagram whose entire credibility is measurement. And one of the five primary texts calls vanity sizing "gaslighting"; in a category where the audience has spent thirty years being told their body is the variable, borrowing clinical vocabulary to make the point is a bad trade for a sentence you do not need.

Ease is the family, and the specific fear it answers is one nobody in apparel advertises to: the day the shirt you have worn for nine years is discontinued. On Instagram that is not a fear at all, because that reader is looking for the next thing. Here it is the reason somebody buys — the fit lives on a profile rather than on a label, so the retirement of a product becomes a non-event. Read the headline set for how consistently the run found that idea: "Your fit doesn't get discontinued", "Your measurements outlive the label", "The shirt changes. Your fit doesn't."

The direction is an email-inbox screenshot and it demonstrates both what that format buys and what it costs. What it buys is a whole relationship in one frame: a standing order, a fit-unchanged confirmation, a reorder that took 84 seconds, a discontinuation notice that is reassuring rather than bad news. What it costs is that a rendered interface has internal rules the model does not follow. The timestamps run 7:58 AM, then 11:47 PM, 11:12 PM, 10:31 PM, 9:41 PM, 8:22 PM, then 6:03 AM and 5:18 AM — an inbox that is not in any order — and the footer claims 9 Unread above a list with one unread dot.

That is a new failure shape worth carrying to the next batch. Three batches established that this pipeline gets rendered sums right and rendered quantities and percentages wrong. This adds ordering and state: any mocked-up interface — an inbox, a chat, an order history, a dashboard — has consistency rules a reader has spent twenty years internalising, and breaking one is the kind of thing nobody can name but everybody notices. Check the screenshot the way you would check a number.

Durability / Reliability is the family, and this is the argument that converts a wardrobe full of good wool nobody wears. The obstacle to buying wool is never the wool — it is the dry cleaner, which turns a garment into an errand. "Machine wash cold, tumble dry low" is a specification about the cloth, not a claim about the reader's life, and it does the same qualifying that a sentence about being too busy to visit a dry cleaner would do, without asserting anything about anybody. It is the clearest small illustration on the page of the rule the whole batch is built on.

Warning-label parody is a good pairing and the execution is better than the brief asked for: the parody structure carries a real care label at the bottom of the frame with correct symbols, so the joke and the specification are the same object. Note also that the four numbered tags — 1, 10, 25, 40 — match the four rows of the logbook beside them. That is the rendered-quantity check that the gym batch's whiteboard failed, and it passes here.

It cannot run as it stands, and it is the cleanest example this cluster has produced of the SaaS Facebook finding that where a direction offers a place to put proof the model fills it. The logbook renders a four-row test series — 0.2%, 0.3%, 0.4%, 0.4% shrinkage, shape RETAINED at every stage — and "0.4% shrinkage after 40 washes" is also one of the five headlines and a sentence in the body. Nobody ran that test. A rendered laboratory result is a stronger claim than a rendered testimonial and a harder one to walk back, because it looks like data. Either run forty wash cycles and publish what happens, or take the logbook out of the frame and keep the joke.

Repeat Customer is the family and wide fittings are the perfect subject for it, because a shoe last that has not been re-cut in twenty years turns a repurchase into a phone call with three numbers in it. "E to 6E" is also the batch's model sentence: it reaches every person who has worn the sides out of every shoe they own, and it says nothing whatsoever about their feet. Compare it with the alternate headline "Built for feet that don't fit trends", which is doing the same job by describing the reader — a small drift, and the direction the whole category drifts in.

The style was pinned as a stat-hero graphic and what came back has no statistic in it at all. The number filling the top half of the frame is a year, 2004, and the proof underneath is a laminated width card and a pair of derbies on Last No. 11. That is a genuinely interesting substitution: asked for evidence, the model reached for a fact about the company rather than a fact about customers, which is the safest thing it could have done. The reference card is the transferable object here — a physical artefact that makes reordering trivial is worth more to this buyer than any review count.

The copy did not stay that safe. The primary text is a first-person customer account — "I've been ordering the same last number since 2009. Haven't set foot in a fitting room since." — from a customer who does not exist, on an ad whose family is proof. That is the same fabrication the Instagram twin's review wall carries, in a quieter costume, and the fix is the same: attribute it to somebody real or write it in the brand's own voice. Two smaller things: the width card shows E, EE, 3E and 4E where the brief and the headline both say E to 6E, and "$255–$345" is an invented range.

Us vs. Them is the family, and the competitor is not another brand — it is the category of advertisement this ad is sitting inside. Facebook has spent a decade training this buyer that a beautifully photographed linen dress is a parcel that arrives in six weeks from an address nobody will print. So the differentiated claim on this app is not about the garment at all: it is that the shop is real. "If the ad won't show you the box, don't order" is a test the reader can apply to every other ad in the feed, including the ones running against you, which is what makes it a positioning ad rather than a defensive one.

The execution is a checklist against a photograph and it is the strongest structural idea in the batch — a comparison chart whose rows are all logistics (return address, delivery, what you see, price) rather than features. One rendering fault: the left column has five rows and the right column has four, so "Proof it's real — none" is answered by an empty cell in the one place the ad most needs an answer. That is a two-minute fix and it is currently the weakest point in the strongest ad here.

And it must not run, for a reason that is exactly the ad's own subject. The parcel in the photograph is stamped 1234 INDUSTRY ST., UNIT 7B, LOS ANGELES, CA 90021 — a real ZIP code in the Los Angeles garment district, on a street name that exists, in an ad arguing that this address is genuine. The SaaS Facebook batch found the same class of fault as a rendered URL on a real top-level domain; this is the physical version, and it is worse, because somebody occupies that unit. Put your actual warehouse address in the frame, which is the only version of this ad that means anything anyway.

Free Trial / Sample is a family apparel almost never uses and this is the case where it is obviously right: a jacket somebody will own for fifteen years, chosen from ten cloths, none of which can be judged on a screen. Posting a swatch card is a real lead mechanic — it needs a name and a postal address, which is a form worth filling in — and it is the reason this is the one Leads ad on a page that is otherwise Sales and Traffic. It is also a mechanic this buyer recognises and the Instagram audience would find absurd, which is a useful test of whether an offer belongs on this app.

Look at how the ad handles price. "£395" appears on a card in the frame and in two of the five headlines, on an ad that is asking for an address rather than a payment. That is the right way round: naming the price before the form filters out everybody for whom the answer is no, so the addresses you collect are worth posting a card to. An offer that hides the price collects a longer list of worse leads.

The defect here is mechanical rather than editorial, and it is the second occurrence of something the SaaS Facebook batch found once. The 4:5 export is not re-composed — the photograph sits between hard horizontal edges at rows 319 and 1034 of a 1350-pixel frame, with flat cream above and below that happens to match the ad's own background, so it reads as generous white space until you measure it. The mill ad further down has the same fault in a more obvious form. Open the re-framed file at full size before you upload it.

Bundle is the family and the audience is the argument: somebody who would rather do almost anything than spend a Saturday in a changing room. A five-piece set is not a discount play for that reader, it is a way to make the decision once. "Measured once. Reordered forever." does more work than the twelve dollars saved, and the ad is right to put the arithmetic on the right-hand side and the promise underneath rather than the other way round.

The numbers check out, which after four batches of rendered-evidence failures is worth stating plainly. $32 plus $32 plus $34 plus $34 plus $28 is $160, which is the struck-through anchor; the set is $148; five garments are drawn against a headline reading five. Sum, anchor and quantity all agree, on an offer-stack graphic — the exact composition that produced the SaaS batch's invented struck-through price. The lesson is not that the pipeline has improved but that a stack built from a list you supplied adds up and a stack built from a discount it invented does not.

One caution and one omission. The struck-through $160 is only honest if the five pieces genuinely sell at those prices individually — a crossed-out anchor is a factual claim about your own pricing, and it is the sort of claim that gets an account restricted rather than an ad rejected. And "Includes your one-time fit consultation" is rendered into the image, which commits you to a service that has to actually exist and be staffed.

Restock is the exact inverse of the Instagram twin's Launch Drop, and putting the two pages side by side is the clearest illustration of the platform axis in this cluster. The drop ad sells a thing that has never existed to somebody who wants to be first. This one sells a thing that already existed to somebody who missed it, and the scarcity is not manufactured at all — a knitting run has a number of garments in it and a sell-through history. "Bracken sold through in eleven days last time" is a fact if you have a ledger, which is why the concept was built around one.

The per-size alert is the part worth stealing. A general mailing list tells everybody that something is back and disappoints most of them; an alert per size tells one person that the thing they wanted is available in the only variant that matters to them. It is also the only mechanic on this page that would justify buying Leads instead of Sales — and the ad does not, because by the time the notification fires the reader is a purchase rather than a lead.

The rendered detail is unusually clean. The lock screen reads "Tuesday, 4 March" and the copy dates the batch February 2025 — and 4 March 2025 was a Tuesday, so the weekday and the date agree with each other and with the story. That is a rendered consistency check passing in a place nobody would have looked. Two things to change before running: "Bracken" is a colourway the model invented and reads close enough to a brand name to be worth renaming, and £215 with "all sizes currently available" is a stock claim that has to be true on the day the ad serves.

Care / Love is the family and gifting is the mode Facebook has and Instagram does not. The insight is sharper than most gift advertising: the failure is not that the present was wrong, it is that nobody could say so. A prepaid exchange card in the box lets the recipient fix the size by post without a conversation, which sells the garment by removing an emotional risk rather than a financial one. Note too that this is the one ad on the page whose copy is about a person — and the person is a third party, not the reader, which is precisely why it stays clear of the policy the rest of the batch is designed around.

The direction is an iMessage screenshot and it is a genuinely effective piece of writing: the tags still on, the drawer, "we tried so hard to pick something she'd love". It is also the batch's most quietly deceptive object. It is a fabricated private conversation between two named siblings about their mother, presented as a real screenshot, and the fact that it is well written is what makes it work and what makes it a fabricated testimonial. The format is not the problem — you can shoot this with two people who agree to it, or write it as an obvious illustration rather than a screenshot — but it cannot ship as a document of something that happened.

Second, smaller, and the same for every screenshot direction in this catalogue: the frame reproduces Apple's iMessage interface closely enough to be recognisable, down to the compose bar and the mic icon. Fine on an article page that labels it a specimen, and a decision worth making deliberately before you spend money running it.

Behind the Brand fills the Founder POV row the Instagram twin left empty, and it answers the objection that page raised: a founder story only works when it ends in a design choice a competitor did not make. This one ends in an object. Two lambswool jumpers, two tabs, the same price and the same photograph — and one of them has a woven tag naming the machine that knitted it and the week it was made. That is not a feeling about fashion, it is a tiebreaker, and it is the one argument on this page that is genuinely aimed at somebody who has not decided to buy anything yet, which is why it is the single Awareness buy.

The photograph is the best in the batch and the crop is the worst. A tag against the cast-iron frame it names, the mill floor soft behind it — and the 1.91:1 export is the square draft inset, with a 56-pixel white bar above the picture and a 46-pixel bar below, not even symmetrical. This is the second padded re-frame on this page and the third this cluster has found. It matters most on 1.91:1, because that shape routes to the right-hand column and search results, which render small: losing a sixth of the height to white bars on the smallest surface on the app is the whole asset.

It also cannot run as written. The primary text is in the first person — "That's the one I bought" — from a customer who does not exist, which makes this the second fabricated testimonial in the batch and the second on an ad whose family is not testimonial. The machine number, the week number, the forty people and the three generations are all invented too, and unlike a price they are not placeholders a reader would recognise as such. The structure is worth keeping exactly as it is: name the frame, name the week, put it on the tag. Then it has to be your frame and your week.